11/15/94
 
  HACMP/6000 and I/O Pacing
 
 
 
  SPECIAL NOTICES
 
       Information in this document is correct to the best of our
       knowledge at the time of this writing.  Please send feedback
       by fax to "AIXServ Information" at (512) 823-4009.
 
       Please use this information with care.  IBM will not be
       responsible for damages of any kind resulting from its use.
       The use of this information is the sole responsibility of
       the customer and depends on the customer's ability to eval-
       uate and integrate this information into the customer's
       operational environment.
 
       +----------------------------------------------------------+
       |                                                          |
       | NOTE:  The information in this document has NOT been     |
       | verified for AIX 4.1.                                    |
       |                                                          |
       +----------------------------------------------------------+
 
 
  ABOUT THIS DOCUMENT
 
       This information is also available in Flash Form as document
       G010493.
 
  I/O PACING AND HACMP/6000
 
       AIX users have occasionally seen poor interactive perform-
       ance from some applications when another application on the
       system is doing large disk writes.  Under certain conditions
       during large disk writes, huge I/O queues are built.  These
       queues can take several seconds to complete.  While the
       write is being completed, an interactive process can be
       severely affected if every disk read spends a great deal of
       time moving up the queue.
 
       Under these conditions, HACMP/6000 cannot successfully get
       keep-alive packets out from the affected node.  The lack of
       a keep-alive activity is interpreted by the cluster manager
       as node failure; and the disk-active node is "failed" by its
       partner.  When the long write finishes, the disk-active node
       resumes its keep-alive activity.  However, its packets are
       now out of sync with the takeover node, which then kills the
       disk-active node with a RESET packet.
 
       I/O pacing allows you to tune the system in a way that more
       equitably distributes system resources during large disk
       writes.  This is done by setting high- and low-water marks.
       If a process tries to write to a file that is at the high-
       water mark, it is required to wait until enough I/Os have
       finished to reach the low-water mark.
 
 
 
 
 
                                            HACMP/6000 and I/O Pacing  1
 
 
                                                                11/15/94
 
  PROCEDURE FOR ENABLING I/O PACING
 
       By default, AIX is installed with high- and low-water marks
       set to zero, which disables I/O pacing.  To set the values
       to something other than zero, thereby enabling I/O pacing,
       choose the following options from the smit menu:
 
          System Environments
            Change/Show Characteristics of Operating System
 
       Then change the value in one or both of the following two
       fields:
 
             HIGH water mark for pending write I/Os per file
             LOW water mark for pending write I/Os per file
 
       While the most efficient high- and low-water marks will vary
       from system to system, an initial HIGH = 33, LOW = 24 is a
       good place to start.  The results from these settings are
       only slightly reduced write times, and consistently correct
       failover behavior from HACMP/6000.
 
       Enabling I/O pacing does slightly slow down very large disk
       writes, but it is REQUIRED for an HACMP/6000 cluster to
       behave correctly during large disk writes.  If you antic-
       ipate large blocks of disk writes on your HACMP/6000
       cluster, it is strongly recommended that you enable I/O
       pacing.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
                                            HACMP/6000 and I/O Pacing  2
 
 
                                                                11/15/94
 
  READER'S COMMENTS
 
  Please fax this form to (512) 823-4009, attention "AIXServ Informa-
  tion".  You may also e-mail comments to: elizabet@austin.ibm.com.
  These comments should include the same customer information requested
  below.
 
  Use this form to tell us what you think about this document.  If you
  have found errors in it, or if you want to express your opinion about
  it (such as organization, subject matter, appearance) or make sug-
  gestions for improvement, this is the form to use.
 
  If you need technical assistance, contact your local branch office,
  point of sale, or 1-800-CALL-AIX (for information about support offer-
  ings).  These services may be billable.  Faxes on a variety of sub-
  jects may be ordered free of charge from 1-800-IBM-4FAX.  Outside the
  U.S. call 415-855-4329 using a fax machine phone.
 
  When you send comments to IBM, you grant IBM a nonexclusive right to
  use or distribute your comments in any way it believes appropriate
  without incurring any obligation to you.
 
  NOTE:  If you have a problem report or item number, supplying that
  number may help us determine why a procedure did or did not work in
  your specific situation.
 
  Problem Report or Item #:               Branch Office or Customer #:
 
  Be sure to print your name and fax number below if you would like a
  reply:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  ______________________________________________________________________
 
  END OF DOCUMENT  (hacmp.iopacing.ha_, 4FAX# 2483)
 
                                            HACMP/6000 and I/O Pacing  3